Now that we know we’re getting Apple Intelligence, and we have GenAI assistants popping up everywhere you look, even in places they shouldn’t, the use of GenAI will extend outward even further, and be used in more scenarios throughout our life, both in professional and personal contexts. This is a big problem that truly requires a well considered solution, because I am a different person when I communicate with an old friend vs. a new prospective client.
This is something that I’ve been thinking about a lot since ChatGPT started remembering things about current and previous conversations, and since the announcement of Apple Intelligence (which I predicted pretty accurately here, if I do say so myself) we’re going to need the ability for our GenAI assistants to identify, learn, and shift tone, voice, language, and personality in sync with the user as their real world contexts change.
Most simply, and most importantly, will be context-shifting between work and personal contexts. We all act, talk, converse and communicate differently between these two worlds, and I definitely do not want to have my GenAI communicate to my wife with the same phrases and language I would use when talking to a client — or vice-versa. That could get really weird, really quickly, regardless of which way you look at it.
Maybe you’re thinking “you would use GenAI to talk to your wife?!?”… and the answer is, probably, yes. I mean, GenAI has moved so quickly, and any and every app you can think of has already begun to, or has already inserted GenAI into it’s workflows — so yeah, I believe this will eventually become so pervasive that everyone will use it for everything, and everyone they communicate with. Eventually, my GenAI agent will craft my message to you, and your GenAI agent will read it and communicate it to you, or simply respond on your behalf. For right now, it seems there’s one voice — I mean you can tell ChatGPT how you want it to respond, what you want it to make you sound like or in what tone of voice, especially if you create your own GPTs, which is a great start — but I want the flexibility to sandbox my personality traits, expressions, colloquialisms, and general tone to groups of people, and even with particular individuals. Again as an example, there is no one else in this world who I communicate with in the exact same way as I do with my wife.
Off the top of my head, I can think of a few key and unique personality contexts, and perhaps constructs, that I would want my GenAI to identify, learn, and use as a unique personality framework when used to communicate to certain people or members of a particular group (in no particular order, of course):
Work:
Personal:
These 11 contexts alone are vastly different. My son is 14, and when I communicate with him, my sense of humour devolves to his level of maturity (note: it’s usually not that much more mature to begin with, so maybe this isn’t the greatest example to use first). My daughter is just about to turn 8, so, there’s a drastically different style there for sure. From a work perspective, I certainly communicate differently with my employees than I do with my clients, and I speak to new clients with a different tact than I do with those we’ve been working with for years. Each of these can further be broken down with unique human to human interaction cues that are specific to each unique relationship sub category, even a specific individual within one of those groups.
As ChatGPT, other isolated app-specific AIs, and Apple Intelligence become even more widespread than they are today, having their responses tailor themselves to the individual or group you are communicating with will become more and more critical for deep user adoption across their entire life, and across their vast and varied social network. This will be a tough one for any company to pull off, other than Apple — though Google with their Android phones, Workspace products, Chromebook devices and various personal media apps like Google Photos are a close second — because the depth at which many of us have jumped into their ecosystems, storing face-identified photos in their photo apps, communications with individuals bound to them through text, email, iMessage, file sharing, and more. Sticking with Apple as the front-runner, for many of us, Apple has a wealth of data on how we as individuals express personal preferences, voice, tone and language choices, levels of humour etc and the subtle differences in which we communicate with each individual we interact with. They can read the text conversations I have had with my son over the last 5 years, or the hundreds of thousands of text messages I have sent to my wife over 20+ years, and they’ll learn a thing or two on how I act, behave and talk with each of them.
The reason I focus on Apple here, and somewhat refer to Google, is that people will not want to set up these personas, nor should they have to add people or groups of people to each persona manually. No one will do that. Ok I would, but I am a nerd who’s willing to put in the effort to get the results I want, or just for the excitement to see if / how well it works. Most people wouldn’t be bothered to define their personality contexts, wouldn’t even know the functionality was there, and wouldn’t take the time to label their connections into the personality contexts they created even if they did know the feature was there and wanted to take advantage of it. This is why it has to be done for us, seamlessly, obscured from the user and run entirely in the background, and then just magically work — spitting out messages and emails in the tone and voice we would naturally use with the particular individuals we’re communicating with at the time. This would require a ton of data to read, and be able to distinguish from one person to the next. It would require emails, messages, maybe even call transcripts to pull out these minute details and differentiations. This is why Apple has the advantage.
Look, I don’t always swear in front of my clients, but when I do, it’s only my most favourite clients.
At the end of the day, I am more than happy to use GenAI everywhere, with anyone. But I don’t want, and can’t afford my GenAI hallucinating and throwing an “I love you” into a message intended for a vendor, just because I end nearly every communication with my family with that line. GenAI needs to begin to understand we’re all chameleons, who are perpetually adapting to our environments, contexts and who we’re communicating with. It’s not enough for personal GenAIs to know Adam vs. Logan as primary users. It has to know the differences between Adam to Leah, Adam to Logan, and Adam to Eikman (wife, son, employee respectively).
We all know GenAI is moving extremely quickly, with a more rapid pace of evolution than any other technology of our time. I think it’s critical that we get this understanding of the multi-faceted nature of humans, and how they interact with various individuals, groups and in varied contexts, scenarios and situations in place as soon as possible — so no one tells their client that they love them. Unintentionally, at least.
Starting something? Maybe you’re building a new product or service from the ground up, or revamping an existing one? We can help.
